New M-ary QAM Transmission Payload System

Transcription

1 r AIAA ICSSC-005 New M-ary QAM Transmission Payloa System Masayoshi TANAKA * Nihon University, College of Inustrial Technology, --, Izumicho, Narashino, , Japan This paper presents a new M-ary moulation satellite payloa system that can provie efficient use of the frequency resource an transmitte power. The system incorporates a superpose M-QAM moulation scheme with a spatial power-combining technology. However, the new system is sensitive to the superposition error. This paper iscusses a configuration for the 64-QAM payloa system an also presents the effects of a specially tailore moulation/emoulation metho an forwar error correction (FEC) coing on the superposition error. It also presents theoretical stuies of the signal transmission characteristics of the system an the investigation results in comparison with the conventional systems. The results show that the power efficiency of the new spatial superpose 64-QAM system is better than that of the conventional system. This unique configuration enables the reliable transmission at higher ata rates with the efficient use of power an banwith. I. Introuction The rapi avances in information technology have le to a growing eman for high-spee access to the Internet, anytime, anywhere, at a reasonable cost to the consumer. A broaban multimeia satellite communications system is a promising network access system because it woul allow us to construct a broaban access system easily an rapily over a wie area compare with other systems, such as fiber-to-the-home (FTTH), CATV, or ADSL. However, before satellite communications systems can be use as a broaban network, we have to resolve several big issues. One is the nee for an economical system esign. We will also have to boost capacity an fin a way to transmit over a ban-limite channel at very high ata rates.. Reucing the output power an the power consumption of the high power transmitters is inispensable for constructing an economical satellite system. In this context, employing M-ary signals woul be effective for broaban transmission with a frequency-ban- limite system. A 6-QAM scheme has been employe in a satellite high-spee ata transmission system. However, an M-ary moulation scheme is generally much more vulnerable to thermal noise an nonlinear istortion from power evices than a QPSK scheme. This means we have to increase the carrier-to-noise power ratio (CNR) an reuce the istortion to achieve a specific transmission quality. Some back-off is require in orer to attain an acceptable level of istortion for a practical power amplifier. However, with back-off, the available power from a power amplifier becomes lower than the maximum an the power efficiency ecreases. This is the main reason an M-ary system has not been wiely use for a power-limite system an is the major barrier to actualizing an inexpensive satellite communications system. Several approaches have been propose to overcome the issue, such as a superpose 6-QAM, a spatial power combine superpose 6-QAM, an the use of a linearize TWTA 4. In this paper, the main focus is place on a 64-QAM signal transmission system, in which nonlinearity becomes a big problem. We propose a new system that, through the use of a superpose moulation scheme an powercombining technology, can provie transmission at higher ata rates with high power efficiency. Then, on the basis of this configuration, we present a theoretical stuy of transmission characteristics uner Gaussian noise an evaluate the system s performance in comparison with the conventional system. The effect of power-combining errors an the variation of system elements on total performance an the allowable errors are also investigate. II. M-ary Moulation * *Professor, Department of Electrical an Electronic Engineering, AIAA Member.

2 M-ary communication (communication using M 64QAM symbols) is an effective way to increase transmission cosωt capacity, since each M-ary symbol carries as much ANT information as log M bits. However, it requires higher S () t transmitte EIRPs or larger receiving antenna iameters AM AM Mo Mo because of its intrinsic sensitivity to noise an interference. In aition, it is more sensitive to nonlinear istortion. Σ HPA 4 Thus, M-ary signaling provies us with aitional means AM AM Mo Mo to achieve the optimal trae-offs between the transmission Linear operation rate, transmission banwith, an transmitte power. 6 Three types of M-ary signaling have been propose. sin ωt They are multi-amplitue signaling (MASK), multi-phase signaling (MPSK) an multi-tone signaling (MFSK) In MFSK, the transmitte power ecreases with M. However, AM moulation the transmission banwith increases linearly with M or exponentially with the rate increase factor k (M= k ). In Fig.. Generation of conventional QAM signal. MPSK an MASK, the banwith is inepenent of M, but the transmitte power increases as M /log M= k /k; that is, the power increases exponentially with the information-rate increase factor k. Hence, we shoul use MPSK or MASK if the banwith is at premium. Quasi-constant envelope moulations, such as QPSK an 8PSK, are power efficient in a single carrier configuration, since they can operate on high power amplifiers (HPAs) riven near saturation. Quarature amplitue moulation (QAM) signaling can be viewe as a combination of amplitue shift keying (ASK) an phase shift keying (PSK) or a combination of two inepenently amplitue-moulate carriers in quarrature. The general QAM signal s(t) is given by 5. s () t x() t cosω t y( t) sinω t = () c where x(t) an y(t) are amplitue-moulate signals The generation of a QAM signal is shown in Fig. an a uniform 64-symbol QAM signal constellation is shown in Fig.. QAM offers better spectrum efficiency for quasi-linear channels. However, conventional QAM is not power efficient since it can only operate on quasi-linear power amplifiers with large output back-off (OBO). In this paper, main focus is place on a QAM scheme an a new type of power efficient 64-QAM system is propose an iscusse. III. Effect of Nonlinear Characteristics of Power Amplifier on M-ary Moulation Wireless communications requires a HPA followe by an antenna for raio wave transmission. In satellite communications, we nee to amplify signals to high power levels, for which high-efficiency amplifiers are esirable. Unfortunately, these amplifiers are nonlinear, an using them to amplify signals causes istortion. AM-to-AM conversion is a phenomenon common to nonlinear evices such as traveling wave tubes (TWT) or solistate power amplifiers (SSPA). At the evice input, any signal envelope fluctuations (amplitue moulation) unergo a nonlinear transformation an thus result in amplitue istortion at the evice output. Hence, HPA operation in its nonlinear region woul not be the optimum choice for an amplitue-base moulation scheme such as QAM. AM-to-PM conversion is another phenomenon common to nonlinear evices. Fluctuation in the signal envelope prouces phase variations that can affect the error performance for any phase-base moulation scheme Fig.. Uniform 64-QAM signal such as PSK constellation. c

3 Conventional non-regenerative repeaters are generally operate backe-off from their highly nonlinear saturation region. This is one to avoi nonlinear istortion an appreciable intermoulation (IM) noise an thereby allow efficient utilization of the system s entire banwith. However, the backing off to the linear region leas to lower output power. Supplie power is severely limite in most satellite communications systems an most power is consume by the HPA. Thus, the inefficiency associate with linear power amplification stages are expensive to bear. Therefore, it is important to raise the power efficiency an lower the power consumption of HPAs for economical system construction. AM-AM AM-PM Fig.. Assume nonlinear characteristics of a typical.5-ghz-ban SSPA. A. Nonlinearity of HPA The conventional 64-QAM system configuration is shown in Fig.. Two inepenently amplituemoulate carriers in quarature are summe after moulation an then high power amplifie. Figure epicts the assume HPA characteristics, which represent a typical.5-ghz ban SSPA. Table summarizes the gain compression, phase shift, an thir IM (IM) at the backe-off conitions, namely, OBO of 0,, 5, an 7 B. Table. Summary of nonlinear characteristics at the back-off points (0,, 5, an 7 B OBO) OBO [B] Gain compression [B] B. Channel Coing The key to achieving error-free communication is the use of appropriate reunancy. Channel coing refers to the class of signal transformations esigne to improve communications performance by enabling the transmitte signals to better withstan the effects of various channel impairments, such as noise, interference, an faing. For practical channels plague by faing an impulses noise, coing can yiel substantial gains an the coe scheme can become significantly superior to the uncoe one. In general, ranom-error-correcting coes are not efficient for correcting burst errors an burst-error-correcting coes are not efficient for ranom error correction. In most practical systems, we have errors of both kins. Thus, in this stuy, a concatenate coing scheme compose of convolutional coing an Ree-Solomon (R-S) coing is use for correcting ranom an burst errors. C. Transmission Analysis of Conventional System Phase shift [eg] IM [B] 0 8 The system moel for the transmission analysis is shown in Fig. 4. It comprises an outer coer, interleaver, inner coer, moulator, emoulator, Viterbi ecoer, e-interleaver, outer ecoer, nonlinear element, an aitive white Gaussian noise (AWGN) channel. Figure 5 shows the effect of this forwar error correction (FEC) coing {R-S (08,88) & R=/ convolution} on the conventional 64- QAM system in linear operation. It is seen that a coing gain of 6 B is obtaine at BER=e-5. The transmission performance of the conventional system was also evaluate uner various nonlinear conitions. To unerstan the effects of nonlinearity an FEC on signal transmission, the BER performance was TX Transmitting ata Outer Inter- Inner Coer leaver Coer MOD HPA R-S- Coe Convolutional Coe RX Outer De-Inter- Viterbi Decoer leaver Decoer DEM AWGN Receiving ata Fig. 4. System configuration for transmission analysis.

4 investigate at the OBO of, 5, an 7 B. Figure 6 shows the signal constellation at the B OBO. The signal constellation on the nonlinear channel is eforme by AM-AM an AM-PM conversion, an the original constellation is no longer maintaine. Figure 7 shows the BER performance versus Eb/N0 for ifferent operating conitions. The performance for linear operation is also shown. It is seen that the BER performance egraes ue to the nonlinearity of the HPA. Even at the B OBO point, BER performance egraes by more than.0 B in Eb/N0, compare with the linear operation. Then, to achieve quasi-linear performance, we nee some aitional means to compensate for the istortion, because FEC coing by itself can not o so. Aitionally, a specially tailore linearize HPA must be employe or the HPA must be backe-off below 5 B. The latter woul cause an output power ecrease an low power efficiency. BER.E+00.E-0.E-0.E-0.E-04.E Eb/N0 [B] without FEC With R-S & R=/ Conv Fig. 5. Effects of FEC coing on 64 QAM in linear operation..0e+00.0e-0 B OBO 5B OBO 7B OBO linear BER.0E-0.0E-0.0E-04.0E Eb/N0 [B] Fig. 6. Deforme constellation of 64-QAM signal ue to nonlinear istortion at.0-b output backoff. Fig. 7. BER performance of 64-QAM signal at three output back-off points on a nonlinear channel. IV. New Payloa System with Spatially Superpose 64-QAM Technology A conventional 64-symbol (M=64) QAM signal waveform can be generate using two quarature-balance moulators as shown in Fig.. On the other han, a 64-QAM signal can also be prouce using three QPSK moulators whose levels iffer by 6 B (0, -6, an B) in a uniform constellation as sketche in Fig. 8. The QPSK signals from the three QPSK moulators have a nearly constant envelope an are less sensitive to AM-AM conversion generate in a nonlinear evice that follows them. Therefore, special attention has to be pai only to the phase istortion prouce by each HPA. A. Configuration The propose new system for 64-QAM signaling is illustrate in Fig. 9. The moulation is performe by superpose technology instea of conventional amplitue moulation. The system incorporates three conventional QPSK moulators (QPSK-,, an ). Their signals are fe to power amplifiers, where each QPSK signal is power-,, L, is ivie into three parallel ata streams amplifie separately. The serial input ata stream ( ) 6 4

5 (, ), (, ), (, ). Then, a ata transformation (, ), (, ) (QPSK- an QPSK-) for Gray coing. The output signals s is performe in front of two QPSKs 4 5 6, s, s from the Q PSK moulators are s s s = r exp = r exp = r exp ( jφ ) ( jφ ) ( jφ ) () where r, φ,( i =,, i i ) are the amplitue an phase for each QPSK signal. The amplifie signal S ans of QPSK- an QPSK- are combine in a irectional coupler as follows. 64-QAM QPSK- Σ QPSK- QPSK- Fig. 8. Principle of superpose 64 QAM Phase array ANT Three QPSKs, QPSK- s φ HPA- S, L, 6 S/P S/P Transform, 4 QPSK-, 5 6 QPSK- s s φ φ HPA- HPA- S S S Coupler Fig. 9. System configuration of superpose 64 QAM with spatial power-combining technology. 5

6 S = + () S S The two output signals S (QPSK-) an S are then combine in a vector-sum manner S = S + with a S spatial power-combining technology. Microwave circuits are other caniates for the power-combining process. But they are complex an have an insertion loss that reuces the transmitte power. On the other han, the spatial power-combining technology enables an efficient power-combining process. Thus, the signal constellation C for 64-QAM is expresse as If setting errors α t,β t C = S + S (4) exist as for gain an phase in the combining process in the transmitter, the resultant transmitting output signal T is istorte an expresse as [ t j t T S + S α exp( β )] = (5) B. Phase Compensation in Moulators As iscusse earlier, FEC coe by itself can not correct the errors cause by nonlinear istortion. To improve transmission performance, a phase shifter is employe to compensate the phase eviation generate in an HPA. In the conventional system configuration, two inepenent signal waveforms are amplitue-moulate an the resultant signal envelope fluctuates. Therefore, it is ifficult to compensate AM-PM conversion. On the other han, in the superpose moulation scheme, the three signal waveforms show a nearly constant envelope. Thus, as shown in Fig.9, it is easy to carry out phase ajustments separately as ϕ, ϕ, an ϕ to cancel the phase rotation. As a result, we can maintain the original signal constellation an can expect transmission performance very close to linear operation. In the new system, 8-level amplitue moulators are not necessary, so that the severe requirements for the moulation are relaxe compare with the conventional 64-QAM system. In aition, the power amplifier following the moulators can be operate near a saturation region because the AM-PM conversion is cancele. This unique feature contributes to improving the power efficiency of the system an also enabling economical an flexible communication equipment. C. Spatial Power Combining for Reucing The Combining Error In the new system, a 64-QAM signal waveform is synthesize by the three separately prouce QPSK signals. The QPSK signals are spatially power-combine by phase array technology. At the receiving sie, the combining errors in amplitue an phase between two complex values S (QPSK-) S an (QPSK-+QPSK-) must be kept small to achieve the same signal constellation C = S + shown in (4) as the transmitting sie. The combining errors arise in the superposition in the transmitter shown in (5) an in the spatial power combing process. The former can be ajuste by harware or software. The latter is epenent on the ifference in the path length an raiation pattern between the two signals, an S. The receive signal R can thus be expresse as S [ α exp( jβ )] [ α exp( jβ )] + n = S + S [ α α exp{ j( β + β )}] n R = S + S (6) t t r r t r t r + where α,β r r are gain an phase errors that occur in the spatial combining process an n is noise signal. To suppress the errors, α r an β r, two sets of two-imensional phase array antenna with the same center position are preferable. S 6

7 D. Moulation/Demoulation Metho for Spatially Superpose 64-QAM If setting errors exist for gain an phase in the combining process, the resultant output signal is istorte as shown in (6) an is illustrate in Fig.0. The propose moulation uses a non-uniform 64-QAM constellation illustrate in Fig., in which r is bigger than that of the uniform one. It is effective to increase the minimum istance separating any two constellation points when is rotate against S. S On the other han, the propose emoulation scheme inclues a function that can estimate the combining error α, β by averaging the receive signals for a known symbol S0 = S,0 + S, 0 transmitte for a short time at the beginning of emoulation. The emoulator ecies the receive symbols base on the moifie signal constellation C calculate by the estimate values as expresse as [ α exp( β )] = S + S j (7) C A moifie signal constellation C is illustrate in Fig.. E. Effect of Gain an Phase Errors in Superposition on Transmission Performance To investigate the effect of the errors, the transmission performance was analyze for various gain an phase errors. Figure shows the BER performance with a uniform constellation an no estimation function in the emoulator uner various gain an phase errors. As a matter of course, BER egraes accoring to the eviation from the ieal conition. The BER performance with a non-uniform constellation an an estimation function in the emoulator was also investigate to stuy the effect of the new moulation/emoulation on vector-sum setting errors. The results are shown in Fig.4. The previously mentione concatenate coing scheme compose of R-S coing an convolutional coing was use in obtaining the ata in Figs. an 4. It is seen from Figs. an 4 that the new system with the non-uniform moulation an the emoulation with combining error estimation is effective in improving BER performance. Fig.0 Deforme constellation of spatial superpose 64-QAM signal ue to combining errors. Fig.. Non-uniform 64-QAM constellation for the propose system (r is increase) Fig.. Moifie signal constellation for emoulation of 64 QAM ( gain error of B an phase error of 0 egrees). 7

8 .0E+00.0E-0 Dp=5eg, Dg=.0B DP=0eg, Dg=.0B Dp= 6eg, Dg=.0B Dp= 0eg, Dg= 0B.0E-0 BER.0E-0.0E-04.0E Eb/N0 [B] Fig.. Effects of gain error an phase error in the power-combining process on 64-QAM signal BER performance with a uniform constellation an no estimation function in a emoulator.0e+00.0e-0 New Sys Dp=0eg, Dg=B New Sys Dp=0eg, Dg=B New Sys Dp=5eg, Dg=B New Sys Dp=0eg, Dg=B Dp=0, Dg=0B BER.0E-0.0E-0.0E-04.0E Eb/N0 [B] Fig. 4. Effects of gain error an phase error in the power-combining process on 64-QAM signal BER performance with a non-uniform constellation an an estimation function in a emoulator V. Evaluation an Discussion The propose system requires three HPAs. However, it is possible to operate them with high efficiency near the saturation region. In the conventional system configuration, the BER performance becomes improves as the back-off is increase. However, the power consumption increases. At the 5-B OBO point, the HPA power efficiency is about one-thir 8

9 of that at the saturation point. From Fig. 7, the performance change versus output back-off is obtaine. As back-off increases, the require Eb/N0 for achieving the specific BER ecrease. Table compares the power consumption between the propose system an the conventional one. The new system using the superpose 64 QAM with spatial power-combining technology can operate on HPAs riven in the saturation region. As a result, the total power efficiency is remarkably improve compare with the conventional system in which HPAs are riven in the backe-off linear region. From Fig.4 an Table, it can be seen that.0 B of gain error an 0 egrees of phase error are allowable in the new system. Therefore, the propose 64 QAM with three-qpsk superposition with a spatial power-combining technology is feasible for a practical system. Table. Comparison of power consumption for several system configurations (BER=e-5) Scheme Conition Eb/N0 [B] HPA efficiency(%) Consumption Power (relative) Conventional 64QAM Spatially Superpose 64QAM B OBO 5B OBO 7B OBO Phase err: 0eg Gain err: B Phase err: 0eg Gain err: B Phase err: 5eg Gain err: B Phase err: 0eg Gain err: B Phase err: 0eg Gain err: B VI. Conclusion This paper presente a new satellite communications system featuring an M-ary QAM moulation scheme an a spatial-power combining technology instea of two amplitue-moulate carriers in quarrature. The system incorporates three conventional QPSK moulators an combines their output signals in a vector-sum manner to prouce a 64-QAM signal. Thus, it can operate on high power amplifiers riven near saturation an relax the severe requirements for the moulation compare with the conventional 64-QAM system. This paper also presente a specially tailore moulator/emoulator an theoretically iscusse the signal transmission characteristics of the system an the investigation results in comparison with the conventional system. Moreover, on the basis of a specific system example, allowable gain an phase errors in power-combining process were erive. The results of this stuy show that the propose system can improve power efficiency remarkably. References A.Franchi,E. Trachtman, L. Christooulies, J. Sengupta,"Multimeia via Inmarsat", Multimeia, IEEE, vol. 6, Issue 4, Oct.-Dec. pp5 9, (999) K.Miyauchi. S.Seki, an H. Ishio, New technique for generating an etecting multilevel signal format, IEEE Trans. Commun, vol. COM-4, pp. 6-67, Feb, (976) M.Tanaka, New satellite communications system using power combine M-ary moulation technology, AIAA ICSSC,AIAA 00-88, April, (00) 4 R.Schornstaet, N.Rozario, C.Hayes, J.Seiner,& A.Katz, "Performance of multi-carrier 6 QAM over a linearize TWTA satellite channel", AIAA ICSSC, AIAA , May, (00) 5 L.W.Couch, Digital an Analog Communication Systems, pp. 5-56, Prentice-Hall (00) 9